EXTREME-VALUE STATISTICS FOR SNOWPACK WATER EQUIVALENT IN THE NORTHEASTERN UNITED-STATES USING THE COOPERATIVE OBSERVER NETWORK

A procedure is developed to estimate extreme-value statistics for snow pack water equivalent (SWE) using historical snow depth measurements a t cooperative observer stations in the northeastern United States. The method specifies ''pseudodensities'' that allow transformation of the statistical distribution of the deepest snowpacks to the distribution of extreme SWE values at a location. These pseudodensities vary accor ding to characteristics of the local snow climatology, the geographic location, and the values of the snow depth data themselves. The perfor mance of a suite of theoretical probability distributions for represen ting the resulting distributions of extreme SWE is also investigated, and it is concluded that five-parameter Wakeby distributions provide t he best representations for the region overall. The results suggest th at previous estimates for extreme SWE values may underestimate the wet test snowpacks at northern and/or higher-elevation locations, but may yield overestimates in warmer portions of the northeastern United Stat es.